In the
 United States Court of Appeals
               For the Seventh Circuit
                          ____________

No. 01-2559
UNITED STATES OF AMERICA,
                                                 Plaintiff-Appellee,
                                 v.

SHOU Z. MEI,
                                            Defendant-Appellant.
                          ____________
            Appeal from the United States District Court
       for the Northern District of Illinois, Eastern Division.
           No. 00 CR 128-1—Ronald A. Guzman, Judge.
                          ____________
 ARGUED SEPTEMBER 18, 2002—DECIDED JANUARY 9, 2003
                   ____________


  Before BAUER, MANION, and ROVNER, Circuit Judges.
   ROVNER, Circuit Judge. Shou Mei pleaded guilty to six
felonies stemming from his role in a conspiracy to com-
mit fraud using counterfeit credit cards, and now chal-
lenges the calculation of the financial loss used to deter-
mine his sentences. Although he conceded stealing nearly
$600,000, the district court determined that Mei intended
to cause a loss of more than $1.9 million. Most of the cred-
it cards involved in the conspiracy were never recovered,
so the district court estimated Mei’s intended loss by
multiplying the average maximum credit limit of the re-
covered cards with known credit limits by the total num-
ber of cards used during the conspiracy. Mei contends
that the method used by the district court to calculate
2                                              No. 01-2559

the loss was not reasonably reliable, and that the ac-
tual financial losses he caused more reasonably reflect the
harm. We affirm.
   The story begins in early 1999, just after Mei was re-
leased from federal custody in Ohio after serving time
for attempted credit card fraud (his second credit card
fraud-related federal conviction, it turns out). Under pres-
sure to make good on his gambling debts, Mei returned
to Chicago to raise cash. His fund-raising scheme involved
having accomplices place “skimmers,” portable data stor-
age devices that capture information from the magnetic
strip on the back of a credit card, in Chinese restaurants,
where corrupt cashiers and servers would swipe the cards
of unsuspecting diners. Mei downloaded the information
from the skimmers onto a computer that he used to pro-
duce counterfeit credit cards and matching identifica-
tion. On May 17, 1999, Mei gave four cards bearing the
name “Xin Chan” to an accomplice, Lau Ming, who tried
to use the cards at a Marshall Field’s department store
in Calumet City, Illinois. The cards did not work and
were confiscated. Mei gave four more “Xin Chan” cards to
another accomplice (and a future government informant)
known only as “Person A,” who attempted the same fraud
at the same store. Again the cards did not work and
were confiscated. Mei continued to produce cards, how-
ever. Ming, along with another future government infor-
mant known as “Person D,” successfully purchased $231
in cigarettes from a Wal-Mart in Bloomingdale, Illinois.
Ming and Person D then attempted to buy a $400 camera
at a nearby K-Mart, but the cards were rejected and con-
fiscated. Later, Ming and Person D obtained over $800
in cigarettes from a Highland, Indiana, Cigarette Dis-
count Outlet using cards bearing the name “Jin Lang.”
Between April and June 1999, Ming eventually charged
over $50,000 with cards supplied by Mei.
No. 01-2559                                              3

   On June 15, 1999, Mei hosted a meeting of Ming, Person
D, and accomplices Chris Huang, Zhou Li, and Nom Tin
Chan. There, Mei introduced a profit-sharing arrange-
ment by which he would distribute to each of them coun-
terfeit credit cards and matching identification, and in
turn each would use the cards to obtain cash or goods that
Mei could trade for cash on the illegal market. Any per-
son able to obtain $12,000 or more would share in the pro-
fits of the scheme. In addition, Mei urged the crew to
place his skimmers in restaurants, and offered to pay
$20 for each stolen credit card number.
  At the same time, Mei was trafficking counterfeit cards
to persons operating in Ohio. One of his Ohio contacts, a
government informant known as “Person C,” bought 30
cards and matching identification from Mei in Chicago,
returned to Ohio, and made approximately $12,000 in bo-
gus purchases. Mei then delivered 30 more cards to Per-
son C in Ohio, who charged another $12,000. On another
occasion, Person C and another informant known as
“Person B” met Mei outside of a McDonald’s restaurant
in Highland, Indiana, where they gave Mei $16,300 for
20 cards and matching identification. Through the end of
1999, Mei sold Persons B and C approximately 130 cards
along with several matching driver’s licenses and at least
two skimmers for about $40,000. And by November 1999,
Person B had obtained over $33,000 with the cards, in part
by taking out cash advances at Indiana riverboat casinos.
  The scheme continued into the following year. In early
February, Mei arranged to deliver 20 cards and matching
identification to Person A. Then, on February 19, 2000, Mei
and Chris Huang ordered $30,000 in computer equipment
from Urban Computer in Chicago, telling the salesperson
that they were starting an internet business. They arranged
to pay for and pick up the equipment the next day. Mei,
Ming, Huang, Zhou Li, Nom Tin Chan, and Person A met
at Mei’s home the next morning. Mei distributed 20 cards
4                                                   No. 01-2559

each to Ming, Li, Chan, and Person A, and at noon they
boarded vans and proceeded to Urban Computer. Mei
remained in one of the vans, but directed Huang to negoti-
ate the remainder of the sale. Although many of the 80
cards did not work, Ming, Li, Chan, and Person A produced
card after card until they found some that did work.
Eventually, Ming charged $11,550, Li charged $8,550, and
Person A charged $10,050. As they loaded up the vans
and prepared to do additional “shopping,” federal agents,
tipped off by Person A, intervened and arrested everyone.
Mei was indicted on charges of conspiracy to commit cred-
it card fraud, trafficking in one or more credit cards,
possessing fifteen or more counterfeit credit cards, plus
three counts of aiding and abetting the use of counter-
feit credit cards. See 18 U.S.C. §§ 2, 1029(a)(1), (a)(3), (b)(1).
Huang, Ming, Li, and Chan were named in the same
indictment as co-conspirators. Persons B and C later told
the government that the object of the scheme was simply
to steal as much as possible by using each card up to its
credit limit or until it no longer functioned.
  Mei eventually pleaded guilty to the indictment with-
out a plea agreement with the government. Sentences for
fraud crimes are based in part on the amount of financial
loss, and at Mei’s sentencing the government urged the
district court to set the loss at $1,918,172.27, which it
argued was a reasonable estimate of the sum of the maxi-
mum credit limits of all of the counterfeit credit cards used
during the conspiracy. The government had to estimate
the total amount of credit placed at risk in this case be-
cause most of the cards involved in the conspiracy were
never recovered, and also because some of the issuing fi-
nancial institutions did not cooperate with investigators.
To determine the average maximum credit limits of all of
the cards, the government first separated the cards with
known credit limits into three categories based on when
or where the cards were used or seized. The first category
No. 01-2559                                                    5

included all cards with known credit limits used or seized
in May or June 1999; the second included cards with known
credit limits seized in Ohio; and the third included those
with known credit limits used or seized on February 20,
2000. The government excluded from its calculation two
cards with unusually high credit limits of over $157,000
and $75,000. The government then multiplied the average
maximum credit limits of the cards with known limits
in each category—$12,066.67, $6,942.86, and $12,864.64,
respectively—by the total number of cards known to have
been used for each category—25, 160, and 34, respec-
tively—and added the results for a total of $1,918,172.27.
This sum increased Mei’s base offense level by 12 under
U.S.S.G. § 2F1.1(b)(1)(M).*
  Pointing out that the estimate was based on a sam-
pling of just 16 percent of the cards employed during the
conspiracy, Mei argued that the government’s loss esti-
mate was too speculative. Instead, he urged the court to
adopt one of what he asserted were two more reasonable
methods of calculating his loss. First, Mei suggested mul-
tiplying the average actual loss to each victim ($2,722.61)
by the number of cards placed at risk (219). This calcula-
tion resulted in an estimated loss of $596,251.59. Alterna-
tively, he suggested multiplying the average loss for the
cards in each category by the number of cards seized or
used in each category, which resulted in a slightly low-
er loss estimate of $555,191.79. Both alternatives would
have increased Mei’s base offense level by ten under
U.S.S.G. § 2F1.1(b)(1)(K). The district court rejected Mei’s


* Mei was sentenced under the guidelines manual effective
November 1, 1998. On November 1, 2001, the Sentencing Commis-
sion deleted § 2F1.1 and consolidated its provisions as amended
with § 2B1.1. See U.S.S.G. App. C, amendment 617. All refer-
ences in this opinion to guidelines sections are to the 1998 ver-
sion of the manual.
6                                                No. 01-2559

suggestions, however, finding that the government’s cal-
culation more accurately reflected the loss because Mei
“intended to gouge his victims for every penny he could
with every card that he had in every way that he knew.”
The addition of the 12-level increase to his offense level
set Mei’s guideline range at 57-71 months. The district
court imposed the maximum term allowed in the range.
  On appeal, Mei argues that the district court’s loss
calculation was unreasonable, and that either of his two
suggested alternative methodologies provide a more ac-
curate and reasonable estimate of the loss. He contends
that neither the sentencing guidelines nor the cases inter-
preting them sanction the averaging methodology applied
by the district court; that even if it were proper to employ
an averaging methodology in this case, the methodology
was too imprecise to render a reasonable estimate of the
loss; and that even if the methodology employed by the
district court rendered a reasonable estimate of the loss,
the factual underpinning of its application—the court’s
finding that Mei intended to use the cards right up to
their credit limits—was clearly erroneous. Mei’s conten-
tions, if correct, would limit his time in prison to 57 months.
The district court’s determination of the amount of loss
is a question of fact we review for clear error, although
the application of the sentencing guidelines is a legal ques-
tion we review de novo. United States v. Higgins, 

, 1074 (7th Cir. 2002).
  The base offense level for fraud and deceit crimes is
subject to graduated increases based on the amount of
financial loss attributable to the defendant, if the loss
exceeds $2,000. U.S.S.G. § 2F1.1(b)(1). The commentary
following the guideline, which is an authoritative interpre-
tive aid explaining how the guideline should be applied,
U.S.S.G. § 1B1.7; United States v. DeCicco, 

,
1537 (7th Cir. 1990), instructs that the amount of loss
considered is the defendant’s intended, rather than actual,
No. 01-2559                                                  7

loss, if an intended loss can be determined, and if the
intended loss exceeds the amount of actual loss. U.S.S.G.
§ 2F1.1, comment. (n.8). In calculating an intended loss, the
sentencing court is required to make only a reasonable
estimate of the loss given the available information. 

   Intended loss, if it can be determined, is applied in fraud
cases because the actual amount of money or property
obtained generally understates the true financial loss.
See id.; United States v. Snyder, 

, 1295 (11th
Cir. 2002). This is so because the purpose of fraud stat-
utes such as the statute Mei violated, 18 U.S.C. § 1029, is
to punish the scheme, not simply the unlawful taking
of money or property. See United States v. Coffman, 

, 333 (7th Cir. 1996). Indeed, the fraud would
have been complete even if Mei had not used the cards.
See United States v. Scott, 

, 552 (7th Cir. 2001).
Thus, in determining an intended loss courts focus on
the amount that the scheme placed at risk, not the
amount of money or property stolen. Id.; United States v.
Bonanno, 

, 509-10 (7th Cir. 1998); United
States v. Strozier, 

, 284 (7th Cir. 1992); see also
United States v. Sowels, 

, 251 (5th Cir. 1993)
(financial loss caused by credit card fraud is the combined
credit limits of the stolen cards).
  Both of Mei’s suggested methodologies estimate the ac-
tual losses he inflicted, so either might stand in as an
acceptable alternative if his intended loss could not be
reasonably estimated. We believe, however, that there
was sufficient information for the district court to deter-
mine the intended loss, and that the methodology it em-
ployed to do so was appropriate. Despite Mei’s contention
otherwise, we, as well as other courts and the Sentenc-
ing Commission, have approved averaging as a rea-
sonable method of calculating intended loss in fraud cases.
See 

; United States v. Egemonye,
8                                             No. 01-2559

, 428-29 (1st Cir. 1995); United States v. Wai-
Keung, 

, 877 (11th Cir. 1997); United States
v. Koenig, 

, 271 (9th Cir. 1991); U.S.S.G.
§ 2F1.1, comment. (n.9); see also United States v. Jarrett,

, 530-31 (7th Cir. 1998) (approving averag-
ing to calculate reasonable estimate of drug quantity in
narcotics trafficking). In a case such as this, where the
total number of counterfeit cards is known but the total
amount of credit placed at risk is not, multiplying the
average credit limits of cards with known limits by the
total number of cards involved in the conspiracy renders
a reasonable estimate of the loss.
  Mei contends, however, that the representative sam-
pling of cards with known credit limits was far too small
to render a reliable estimate of the average maximum
credit limits in each category, so that the district court
lacked sufficient information to determine an intended
loss. The respective averages of the three groups of cards
were extrapolated using roughly 12, 8.75, and 82 percent
of the cards. Although we have sustained a loss deter-
mination based on a much smaller sampling, see 

(2 of 200 cards), what matters is that the
representative sample yield a reasonably reliable figure
under the circumstances, and that is the case here. The
average for the cards used or seized on February 20, 2000,
was $12,864.64, a figure extrapolated from 82% of the
cards. Yet the average determined for the cards used or
seized in May and June 1999 was just slightly under this
number, $12,066.67, despite being based on a sample
of only 12% of the cards. And the average extrapolated
from the cards seized in Ohio, which was based on
only 14 of 160 or 8.75% of the cards, was significantly
lower, $6,942.86. Thus, the categories containing the least
amount of information rendered averages more favor-
able to Mei than the category with the best informa-
tion. Considering also that two cards with unusually
No. 01-2559                                                9

high credit limits were eliminated from the mix, we be-
lieve that the district court’s estimate of the averages
for each category was sufficiently reliable, if not conserva-
tive.
  Of course, these numbers hold water only if, as the
district court found, Mei “intended to gouge his victims
for every penny he could with every card that he had
in every way that he knew.” Mei argues that this finding
is contrary to the evidence before the court at sentencing.
He points out that only one card was used up to its maxi-
mum credit limit, and that only two other cards were
used close to their maximum credit limits. He further
argues that he could not have intended to use each card
up to its limit because it was not possible for him to do
so. But Mei’s failure to exploit the cards to their fullest
potential is not dispositive—it is irrelevant that the
full extent of the risk did not materialize. 

; 

; 

; United States v. Studevent, 

,
1561-62 (D.C. Cir. 1997). The amount of money or prop-
erty Mei was able to obtain from the counterfeit cards
was limited not by his own restraint, but by facts beyond
his control. Mei did not know the limits of each stolen
credit card number when he obtained it—for all he knew,
a skimmed card could have a credit limit of $150 or
$150,000 (and one did in fact have a limit of more than
$150,000, although it was excluded from the calculation).
Some cards when stolen may already have been at or
near their limit, so that they did not work when Mei
attempted to use them. Or, some of the cards did not
work because their rightful owners were delinquent in
paying a bill or had previously reported the number stolen.
Or, Mei was arrested before he could more fully exploit
the scheme. That he was thwarted from extracting the
full potential of the cards by events beyond his control
does not speak to whether he intended to do less or was
less culpable. See 

.
10                                            No. 01-2559

  Here, there was plenty of evidence for the district court
to conclude that Mei intended to use each card up to its
limit if possible: the profit-sharing arrangement Mei
introduced, whereby his accomplices would share in the
profits of the scheme only after they had charged at least
$12,000, suggested to and encouraged his accomplices to
charge as much as possible; Persons B and C indicated
that they purchased cards from Mei with the understand-
ing that they were to use the cards to their maximum
limits; and there is nothing in the record or the briefs
suggesting that Mei or his accomplices set a limit on how
much they could use an individual card.
  Because we see no error in the calculation of Mei’s in-
tended loss, we AFFIRM the judgment of the district court.

A true Copy:
      Teste:

                       ________________________________
                       Clerk of the United States Court of
                         Appeals for the Seventh Circuit




                   USCA-02-C-0072—1-9-03